[Maxer]

My scene was set up with Houston being 50' above sea level in which case I just raised the entire ground plane 50' in the Y direction just as I did for the San Francisco example that I posted earlier. In both cases changing the elevation did noting to affect the pink in the scene. I'm not sure what your trying to tell me, do you believe the pink sky is a result of something I'm doing because I don't see how that is possible.

[superbad]

I decided to test this too- it seems all these tests with buildings and mountains are confusing the issue. I went to Flickr and got a photo of the sea- the sea is flat, at sea level, and has an unobstructed view of the horizon. And Flickr displays EXIF data, so we can setup the camera exactly the same.

Here is the photo I chose, Swansea beach (Wales), 2:00pm in June:
http://www.flickr.com/photos/richardandgill/159833378/


I copied the camera settings exactly, except I set the focal length to the 35mm equivalent, and left the film size at 35mm. I set the location to Cardiff instead of Swansea (close enough). Judging by the map provided with the camera data, and the fact that the sun is not visible, I estimated the shot was taken toward the southeast. The scene contains nothing but a 1km groundplane. The only dodgy thing in the EXIF data here is the exposure compensation (15 EV), which is clearly either wrong, or being expressed in some non-standard way. The Maxwell output was overexposed compared to the photo, so I reduced the exposure 1 stop (ISO 100 -> 50).

Here is the result:

[Maxer]

Your Flickr image is exactly what I would expect a sky during the day to look like and this is what I think we should be getting from our tests if indeed the sky is functioning properly. I'm still not sure why we are now being told we need to compensate for sea level as well as know what the different Trubidity, Ozone, and Water setting should be for each location we are working in....shouldn’t those settings be automatic?

I'm about to call it a night but I'll definatley be checking this thread in the morning, I hope we can figure out what is going on tomorrow.

[Tyrone Marshall]

Superbad,

I can show you images of San Francisco near water and so on that have this coloring as well (granted these are not my personal images, these come from the internet):





Besides, Juan has already responded that there will be an option to adjust the sky in this fashion:

Thanks a lot for such interesting comments.

This sky model asumes the observer is always located at altitude 0, and this is why we always see the pink-orange gradient in the horizon. Soon we will add an altitud parameter to control it. Besides, it is important to take in consideration what happens in real life, the ground is never perfectly flat, otherwise we always would see the red line in the horizon if we are at altitude 0. It is not common at all, so a scene like the one Tyrone has showed is more correct to avoid to see always below our own feet.

Regards,

Juan

[Tyrone Marshall]

Your Flickr image is exactly what I would expect a sky during the day to look like and this is what I think we should be getting from our tests if indeed the sky is functioning properly. I'm still not sure why we are now being told we need to compensate for sea level as well as know what the different Trubidity, Ozone, and Water setting should be for each location we are working in....shouldn’t those settings be automatic?

I'm about to call it a night but I'll definatley be checking this thread in the morning, I hope we can figure out what is going on tomorrow.

Devin,

I have one more question for you.

So version 1.1 and 1.0 rendered these sky scenes correctly? I mean the previous versions of Maxwell Render did not have this coloring that you and others dislike, at all at the horizon line?

[VALKAMA]

This thread is almost to the point of being silly becasue of you Tyrone. What are you trying to prove? Its obvious the skys are much more pink in version 1.5 than in previous versions. And please dont try and convice me that the sky is always pink like this because it is not. Sometimes yes the sky is pink and sometimes its red orange yellow blue black. Maxwell is simply showing way too much pink in the sky with this release and its obvious by all of the latest postings. The previous skys were great and felt the most natural. I will be happy if we have some control over the sky color as currently i dont want to show pink skys to my clients

[Tyrone Marshall]

This thread is almost to the point of being silly becasue of you Tyrone. What are you trying to prove? Its obvious the skys are much more pink in version 1.5 than in previous versions. And please dont try and convice me that the sky is always pink like this because it is not. Sometimes yes the sky is pink and sometimes its red orange yellow blue black. Maxwell is simply showing way too much pink in the sky with this release and its obvious by all of the latest postings. The previous skys were great and felt the most natural. I will be happy if we have some control over the sky color as currently i dont want to show pink skys to my clients

No one is kidding, I just have observations that the sky is not as much as is being written to be in this thread with respect to coloration. Please disregard the mapping of the second image. The topic here is really about the sky quality and coloration.

I think it not very much different at all. If you can show me otherwise, I would agree but with these samples I cannot:

Here is Maxwell Render version 1.0 - (4/16/2006):



Here is Maxwell Render version 1.5.0 - (5/27/2007):



With these, I rest my case as I really cannot see what is so much more different except a general color cast, maybe that is what you are referring to. If it is not, then I have no idea what else it is as these two images were done at two different times at more than a year apart with two different versions of Maxwell Render and the physical sky is not that much different.

[Hybaj]

Now this is going to be a very interesting debate and I knew that the dev team and testers are going to defend the current model at all costs.

I have to state that I beleive that the formula for the skydome colour might be a quite good aproximation but we're talking about it producing outputs under some perfect fixed conditions. And there are NO perfect conditions in real life. The formula simply doesn't take into account alot of other factors.

You can say unbiased and physically correct and other blablabla but we're dealing with a not that difficult formula.

I paint skies since I was a kid and most of the "pink during day" happens in the winter time. Of course it also happens during summer but even when it does happen it's not looking like what maxwell creates because for example the pink might be too strong or the blue colours will be totaly different. Yesterday I browsed hundreds and hundreds of my own photos created in a probably 6 year span and non of them was atleast 80% close to what I see happening in Maxwell. The pink was either much subtle, the gradient between pink and blue was lighter or the blue colours didn't match very well.

Does it mean that Maxwell is totally wrong?? No but what are such skycolours good for when they don't ussualy match the reality?? This is the dillema folks

You can bash me all you want but it is very very important to look at other softwares which deal with skycolours for years. Vue, Terragen, Bryce and others. Their methods are time-proven. Of course you can see some skies with pink on bottom but not much and not that strong and saturated. People ussualy tend to create different sky conditions and outputs of such actions look much more eye-pleasing - of course not everytime.. it depends on the artist, which is important.

So our problem is that we cannot control the sky like we want. Only few variables and the formula does the rest of the job. Now we're stuck with a formula that creates a very characteristic sky. Not the more "neutral" looking like before which satisfied most of the people.

So what's the solution??? Well you can change it the model before, give us more variables so we can play with the looks (start with the elevation), or play a dead NL bug. I think the best would be to be able to manualy override the formula's solution through wavelength gradient controls.

Just look at other softwares which have much more experience in this field and get inspired

Tyrone : Yes the colours are looking better in the 1.5. But the colours during mid-day not

[Tyrone Marshall]

... it is very very important to look at other softwares which deal with skycolours for years. Vue, Terragen, Bryce and others. Their methods are time-proven. Of course you can see some skies with pink on bottom but not much and not that strong and saturated. People ussualy tend to create different sky conditions and outputs of such actions look much more eye-pleasing - of course not everytime.. it depends on the artist, which is important.

So our problem is that we cannot control the sky like we want. Only few variables and the formula does the rest of the job. Now we're stuck with a formula that creates a very characteristic sky. Not the more "neutral" looking like before which satisfied most of the people.

So what's the solution??? Well you can change it the model before, give us more variables so we can play with the looks (start with the elevation), .....
Just look at other softwares which have much more experience in this field and get inspired

Tyrone : Yes the colours are looking better in the 1.5. But the colours during mid-day not

Now Hybaj,

This is best part of the discussion yet. I think now we are getting to the thick of the real discussion. Are there any samples of some TG2 that you can show where we can compare and contrast? You bring up some great points about choices and such and I think these are worth consideration to continue discussion. I would like to take a good look at how maybe TG2 takes its methodology to sky simulation.

Great stuff and finally after 3 pages, yeah! This is the kind of stuff this forum used to get into a couple of years ago. Lets keep going! Who knows what could come of this deepening discourse?

[juan]

Hi Hybaj,

Now this is going to be a very interesting debate and I knew that the dev team and testers are going to defend the current model at all costs.

It was not my intention. Actually I just pointed out a limitation of the current model regarding altitude that we'd like to improve soon. There were issues in v1.1 sky in both color and intensity and we have further enhanced that area. If now that this items have been fixed we notice that new parameters are needed we will work on that. Thanks again for the interesting discussion here.

Regards,

Juan

[Hybaj]

Yes this might be a interesting debate but will it be an effective one?? So let's move onto some rational debate. I want to put all my ideas into this post.

So the problem is that thousand of different atmospheric conditions create thousands of different skies. As much as I like all the settings for skies in different softwares I feel there's something always missing.

Ok so what are the advantages of other softwares? Let's look at TG2 (tech preview.. free for download). The settings are looking that they have NO "physical corectness" in them. Example

Blue sky density
Blue sky horizon colour
Blue sky additive
Redsky decay
Bluesky exp height
Ceiling adjust
Celing
Floor
Ambient
Fake dark power
Fake dark sharpness
and etc.....

So there is alot of variables which from the first look do not make much sense. Are they physically correct?? I personally have no idea but they do seem working together in some kind of a physical aproximation model. Correct or incorrect as it all might seem you can always create skies that match photos quite well.

Ok so that means that the more options you have the more different outputs you can create. Nice but it also takes more time to tweak the controls to get what you want and some lazy Maxwell users would start arguing that they bought Maxwell because they were fed up with Mental ray and etc because they had SO many settings.

Terragen has no easy parameters like longitude/latitude and date/time. Which I think is a bit of a downside.

So let's combine stuff from the both worlds of Maxwell and Terragen

And this is how it would look like according to my personal preferences which absolutely in no way represent preferences of the whole Maxwell community.

1. Set Long/Lat, Time/Date and hit render. The formula spits out a solution which you can see in the preview window. If you don't like the solution you can move onto step 2.

2. You enable advanced sky parameters which lets you tweak the sky even further. You can tweak it even to unrealistic extremese which might come in handy for people for non architectural work.

3. You hit render and are satisfied with the result




Note : Also the pink color also affects the whole lightning in the picture. So when you put an actual image into background you get a bad composition.
Note nr.2 : Sorry for the clouds in the preview picture

[Becco_UK]

Nuno Faria: That link to the climatology site is very useful thanks.

[giacob]

i would like to point out that most , if not all, of the photos showed from tyron have the point of view very high or high above the horizon... i think if the point of view would be at man height.. no pimkish sky ad midday

[Nuno Faria]

Nuno Faria: That link to the climatology site is very useful thanks.

Glad you liked it

[numerobis]

i would like to point out that most , if not all, of the photos showed from tyron have the point of view very high or high above the horizon... i think if the point of view would be at man height.. no pimkish sky ad midday

with maxwell you are always high above all...